Models

Anthropic Researcher Claims Claude Proved the Riemann Hypothesis

Levent Alpöge, a researcher at Anthropic, announced he used Claude to prove one of mathematics' most significant unsolved problems, though the company has not officially confirmed the breakthrough.

·3 min read
Anthropic researcher claims to have proven Riemann hypothesis using Claude
Anthropic researcher claims to have proven Riemann hypothesis using Claude

A mathematician working at Anthropic PBC says he has leveraged Claude to solve the Riemann hypothesis, a problem long regarded as the central unsolved question in mathematics. Levent Alpöge made the announcement on his personal website, though Anthropic itself has not released any official statement. The disclosure comes just one day after Anthropic's initial public offering on the Nasdaq, which generated proceeds exceeding $93 billion.

In a four-page writeup, Alpöge stated: "I believe I have, working with Claude, discovered a proof of the Riemann hypothesis. What I saw during the course of progress so amazed me that I believe it is one dawn across science of something far greater."

Understanding the Riemann Hypothesis

Bernhard Riemann put forward the hypothesis in the middle of the 1800s, building on the Riemann zeta function—a mathematical concept developed over a century before. Similar to how a software function processes input data to generate output, the Riemann zeta function accepts data and produces results. Both its inputs and outputs take the form of complex numbers, which extend the properties of ordinary numbers with additional mathematical characteristics.

When plotted visually, the inputs and outputs of the Riemann zeta function create a three-dimensional spiral pattern. The hypothesis concentrates on a particular portion of this spiral where the function's outputs all equal zero. The conjecture posits that every input producing a zero output must contain the value ½.

The hypothesis holds tremendous significance in mathematics for several reasons. Despite addressing a single function that is not inherently complex, its consequences reach far beyond—particularly to the behavior of prime numbers. Should the hypothesis prove true, what appears to be randomness in prime number distribution would actually follow an underlying pattern.

Additionally, hundreds of mathematical theorems rest on the assumption that the Riemann hypothesis is correct. Proving it would simultaneously validate all these dependent theorems.

The Scope of Claude's Proof

According to Alpöge, Claude demonstrated not just the classical Riemann hypothesis but also its generalized form. The generalized variant extends the hypothesis beyond the Riemann zeta function to encompass certain related mathematical functions. Alpöge suggests that Claude may eventually be capable of proving an even more expansive iteration of the hypothesis.

The proof generated by Claude spans more than 250,000 pages, which Alpöge has not yet made public. However, he has released a version encoded in Lean, a programming language designed for verifying mathematical proofs. This Lean implementation contains millions of lines of code.

Proofs written in Lean function as executable software programs. When a Lean program executes successfully without errors, the mathematical reasoning it encodes is guaranteed to be valid. This property makes Lean proofs particularly valuable for the mathematical community, as it streamlines the verification process for newly discovered proofs.